Cognivix

About the offering

Cognivix is an early-stage industrial robotics software company founded in May 2025 and incorporated in Delaware, headquartered in San Francisco, CA. The company is raising on Republic via a SAFE instrument with a $7M valuation cap (reduced from $12M under special terms), with a $100 minimum investment and an October 1, 2026 deadline. Core Product: AiKit is described as a zero-integration operating system that enables industrial robotic arms (compatible with FANUC and ROS-based platforms) to learn new assembly tasks from a single human demonstration, without writing any code. The workflow has three steps: (1) a human demonstrates the task once, (2) AiKit uses the demonstration combined with 3D CAD models and physical AI to generate a manipulation sequence with sub-millimeter precision, and (3) the task is deployable across multiple robots within five minutes. A DevKit (compact robotic development kit) serves as the entry point, bundling three months of AiKit software, then converting to annual per-robot licenses priced at $10,000 (or $7,500 after discount/reseller fee). Market Opportunity: The company cites a $3 trillion global market for automating manual assembly, with a $400 billion serviceable segment focused on sub-millimeter tolerance, rigid-part, and multi-step manipulation. The macro driver is that up to 80% of production remains manual, traditional automation is too rigid for small-batch/on-demand/onshore manufacturing, and 95% of manufacturers plan to invest in AI or ML by 2030. Traction & Financials: Cognivix reports $2M in total investments and grants ($1M non-dilutive grants, $1M equity funding through its European affiliate). It has completed $170K in paid proof-of-concept engagements. The company is an NVIDIA Inception Program member and was named one of '20 Manufacturing Startups to Watch' by StartUs Insights. AiKit has been showcased alongside NTT Data and FANUC at trade fairs. Collaborations include FANUC, NTT Data, Technical University of Munich (TUM), Accenture, and Deloitte. Roadmap: The 12-month plan targets shipping AiKit, deploying 200 licenses, and reaching $1.5M ARR. By 2027, the company plans to launch the NTT Data global channel; by 2028, introduce a higher-value 'AiCell' product; by 2029, target $150M ARR and 6,000 robots deployed. The long-term model framing is 800 accounts x 20 cells x $25K/year. Business Model: Software-led per-robot licensing (SaaS/licensing), with enterprise sales driven by NTT Data and hardware opportunities expanded through FANUC. The company aims to reduce custom engineering per deployment over time, handing integration work to partners and integrators. OEM channel partnerships (e.g., aluminum extrusion press manufacturer) are another distribution vector. IP: The company holds a U.S. provisional patent application covering the core learning process (single demonstration to robot-agnostic assembly program). Additional components are protected as trade secrets while further filings are evaluated. Long-term IP strategy includes accumulation of robotic execution data (vision, force/torque, contact) from deployed systems. Team: Co-founders include Daniele Bernardini (CEO, physicist/serial founder, former Siemens global data-warehouse lead architect and TUM lecturer in computer vision/ML for robotics), Francesco Scalise (CRO, serial founder with manufacturing sales background and experience at Lamborghini, Saint-Gobain, Caterpillar), and Giorgio Calapa (COO, MBA with 15+ years in enterprise software at Citi, BofA, Morgan Stanley, Bloomberg). The CTO is Lukas Dirnberger, Chief Scientist is Boyoung An, and Head of Engineering is Suraj Deshini. The advisor bench includes a YC-backed founder with two exits, a former VP of Engineering at iRobot and Symbotic, and a SRI product expert featured in Wired and NYT. Key Risks: Limited operating history (Cognivix Inc. founded May 2025, 1 employee listed); the acquisition of its Italian sister entity may not close on expected terms; advances in foundation robotics models (VLA, generalist AI) could reduce need for specialized manipulation software; capital raised may be insufficient to sustain the business plan; enterprise sales cycles are long.

FAQ